Consistent rewards outperform variable rewards in habit formation

Working paper
Asaf Mazar, Marissa Sharif, and Angela Duckworth

Conventional wisdom suggests that habits form best in response to variable rewards rather than consistent ones. We challenge this view using large scale field data (N > 1M posts) and six pre-registered experiments. In the field and in the lab, consistent rewards promote long-term persistence more effectively than variable rewards. Experimental evidence suggests that this persistence reflects automatic habitual responding. These findings question long-standing assumptions and highlight the role of predictability in habit learning.

Manuscript title: Consistent rewards outperform variable rewards in habit formation.